Perhaps no subsector of the tech industry has been more affected by advances in AI than data, and we’re nowhere near the peak of what the technology can do.
AI relies on data for the information it needs to use to avoid hallucinations and bad outputs. That’s where master data management (MDM) comes in, ensuring that organizational data is clear and accurate. Ben Werth, CEO of MDM firm Semarchy, told IT Brew that it’s essential for both data and AI businesses.
“Customers want to understand what your AI is doing and how you’re coming up with things,” Werth said. “You cannot sell them AI tooling unless they understand how it works—master data management is the foundation for making sure that you have good inputs.”
Trashcans. Garbage in, garbage out may be a cliché, but that could be for a good reason. Syndigo Chief Product Officer Tarun Chandrasekhar told IT Brew that increasing accuracy in the AI space is a direct result of better MDM practices. Those practices, in turn, are informed by what customers demand from vendors.
“We are at a stage where 95% to 98% accuracy is pretty much the requested standard for most of us,” Chandrasekhar said.
It pays off: AI is being increasingly deployed with speed as advancements move from a months or weeks timescale to a matter of days.
“You can actually start working faster on building tools that help you work faster on getting AI to work,” Chandrasekhar said. “This is the self-fulfilling prophecy, in a way, but that acceleration has been something that we’ve never seen before.”
MDM is only going to continue to be important, Werth said, as the importance of good data continues to grow. Companies are prioritizing the need to ensure the inputs match the outputs they want. In other words, “If you do not have the foundation right, all the downstream things are going to fail,” Werth told IT Brew, making the industry essential.
Top insights for IT pros
From cybersecurity and big data to cloud computing, IT Brew covers the latest trends shaping business tech in our 4x weekly newsletter, virtual events with industry experts, and digital guides.
“I think there is a reawakening of what I’ll call the old MDM market,” Werth said. “People are getting more interested and more excited about understanding the importance of good data; it’s becoming a clear competitive advantage, and that is great for the MDM space.”
Trouble brewing?
As with any tech improvement, MDM comes with risk. The threat surface is expanded by its use, Chandrasekhar said, and it’s not only cybersecurity but also privacy compliance and regulations in general. The existing series of checks and balances that exist are in place to deal with the industry as it was before AI, which adds another layer of complication. Finally, there’s the continued need for human control and oversight—something that is going to have to change as AI models are able to take charge of some of their own processes due to the large amount of information at play.
“Humans can absolutely do it, they’re just going to take longer to do it,” Chandrasekhar told IT Brew.
Value proposition. Ultimately, companies need to look at MDM not as a program that can be run and forgotten about, humming along in the background, but as an ongoing process, Chandrasekhar said. It’s more important than ever to ensure MDM works with AI to extract value from data.
“For the first time in my 20 years, MDM is getting much closer to the shelf, where people are like, ‘I can tie my success as an organization directly to how I’m wrangling and monetizing data,’” Chandrasekhar said.